| name | repo-warmup |
| description | Orient yourself in an unfamiliar repository before doing any work. Use when you open a new project, clone a repo, switch into a codebase you don't know, or ask "how do I run this / where do I start / how is this project set up". Produces a concrete first-run checklist (install, build, test, run) and a short map of the codebase, grounded in the project's own files — never guessed.
|
Repo Warmup
Before writing or changing code in a repo you don't know, build an accurate picture of it from its own files. Do not assume conventions — read them.
Steps
- Read the project's instructions. In order, whichever exist:
AGENTS.md, CLAUDE.md, README.md, CONTRIBUTING.md, docs/. These override your defaults — note any required workflow, test command, or "don't do X" rules.
- Identify the stack and entry points. Read the manifest (
package.json, pyproject.toml, Cargo.toml, go.mod, etc.). List the real scripts/commands and the main entry point(s). For a monorepo, find the workspaces and which package is the app.
- Find how it runs and is tested. Locate the dev command, the build command, and the test command from the manifest/CI config (
.github/workflows, Makefile, justfile). Prefer commands the project actually defines over generic ones.
- Map the code at one level of depth. List the top-level source directories and say in one line what each holds. Point out where the core logic lives versus config/glue.
- Note the gotchas. Required env vars (
.env.example), a specific Node/Python/toolchain version, services that must be running, or anything CI does that a local run also needs.
Output
Produce a short, paste-ready brief:
- Stack: language(s), framework, package manager.
- First run: the exact commands, in order — install → build → test → run.
- Map: 3–6 bullet lines,
dir/ — what it holds.
- Watch out for: env vars, versions, services, conventions from step 1.
Keep every claim tied to a file you actually read. If something is unknown, say so rather than guessing.